关键词: PM_2.5, 时空特征, 旋转经验正交函数（REOF）, 地理探测器

Abstract: Smog events are widespread in central and eastern China. Scientifically identifying the spatiotemporal variations and the drivers of PM_2.5, the primary cause of smog, is of great significance for regional prevention and control. Using PM_2.5 data from December 2014 to November 2019 of the national air quality monitoring stations, the spatiotemporal characteristics and variations of PM_2.5 pollution were analyzed by using the methods of geostatistics, REOF and Geodetector. The region classifications were identified, and the driving factors for differentiation of PM_2.5 were identified in each region. The results showed a steady downward trend in mean annual PM_2.5 concentration in central and eastern China, with a mean annual decrease of 3.2 μg·m^-3. The area with high annual PM_2.5 concentration showed a rapid spatial contraction. There was an insignificant increasing tendency of PM_2.5 pollution days in some areas. The concentration of PM_2.5 decreased outwardly in the center of BeijingTianjinHebei, with a characteristic of “higher concentration in the north than in the south”. The strong spatial agglomeration of PM_2.5 concentration was distributed in central and eastern China. The values of global Moran’s I in all four seasons exceeded 0.70 (P<0.01). Hot spots were mostly distributed in and around the North China Plain, while cold spots were distributed in south China. Based on the result of REOF, three regions were classified. Their boundaries roughly overlapped with the topography and geomorphology. The time coefficients showed a “pulse” downward trend, indicating a significant improvement trend of PM_2.5 pollution. Meteorological factors and human activities are essential drivers for the differentiation of PM_2.5 in different regions, with the key drivers being region-specific.

Key words: PM_2.5, spatiotemporal characteristics, rotated empirical orthogonal function (REOF), Geodetector.